You are on page 1of 30

!"#$%& ()*+*,-.& &/ (& /01#"2& 13#+#425$& ,-#+ 6 .

/ 1/2&++&
























W 1010

789:;<=>< =?@@AB?98 C8= ?889DCE?98 F9;?G?<B :?EH?8
EH< &AI9F<C8 I<>?98BJ 3HC;;<8><B KCB<= 98 I<G<8E
<LF?I?GC; <D?=<8G<

Corlnne AuLanL-8ernard, Murlel ladalro, nadlne Massard

Mal 2010








u
o
c
u
m
e
n
L
s

d
e

L
r
a
v
a
l
l

|

W
o
r
k
l
n
g

a
p
e
r
s

h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
















!*/& !I9AF< =)*8C;MB< <E =< /HN9I?< 1G989L?OA< ,M986.E 1E?<88<

93, chemln des Moullles 69130 Lcully - lrance
1el. +33 (0)4 72 86 60 60
lax +33 (0)4 72 86 60 90

6, rue 8asse des 8lves 42023 SalnL-LLlenne cedex 02 - lrance
1el. +33 (0)4 77 42 19 60
lax. +33 (0)4 77 42 19 30

Messagerle elecLronlque / Lmall : gaLe[gaLe.cnrs.fr
1elechargemenL / uownload : hLLp://www.gaLe.cnrs.fr - ubllcaLlons / Worklng apers


h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0

Knowledge diffusion and innovation policies within the European regions:
Challenges based on recent empirical evidence

Corinne Autant-Bernard, Muriel Fadairo
*
, Nadine Massard

Universit de Lyon, F-42023 Saint Etienne, France
CNRS, GATE Lyon-St Etienne, UMR n 5824, 69130, Ecully, France
Universit de Saint-Etienne, Jean Monnet, F-42023 Saint Etienne, France



ABSTRACT

This article builds upon empirical results concerning localised knowledge spillovers to highlight some policy
implications within European regions. The analysis emphasises the role of regional innovation policies in
supporting the institutions that generate knowledge and learning. However, the variety of regional features
presented in the empirical literature suggests that the search for universal policy tools is unrealistic. From this
perspective, we argue that original strategies must be generated to cope with the various dilemmas faced by
regional innovation policies. Such specific strategies require accurate knowledge of local features. Improving
data and indicators to diagnose and monitor regional innovation is therefore presented as a key issue for policy
makers.

JEL Classification: O38, C12.

Keywords: innovation policy, localised knowledge flows, European regions, knowledge-based economy.

Mots cls : politique dinnovation, flux de connaissances localiss, rgions europennes, conomie fonde sur
la connaissance.


1. Introduction

In his 1994 article, Metcalfe identifies two main profiles in technology policies: i) those that take
the possibilities of innovation as given, and thus seek to stimulate innovation by reducing the cost of
R&D activities or by increasing the profitability of private innovation; and ii) those that seek to
expand the opportunities for innovation.
This paper, based on the results of the empirical literature devoted to knowledge spillovers,
locates itself in the second policy perspective. This supposes that innovation policy must go beyond
providing R&D subsidies. We argue that, in a knowledge-based economy, taking into account the
sources, paths and the underlying mechanisms of knowledge diffusion, the primary role of
innovation policy is to create a variety of mechanisms to facilitate the capture and assimilation of
local and external knowledge.

*
Corresponding author at: GATE Lyon Saint-Etienne - CNRS, Universit Jean Monnet de Saint-Etienne, 6 rue basse des rives, 42 023
Saint-Etienne cedex 2, France. Phone: +33 (0) 477 421 963. Fax: +33 (0) 477 421 950.
E-mail address: fadairo@univ-st-etienne.fr (M. Fadairo).
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
2

One major finding of the research field devoted to innovation is that technological knowledge is
not only an output of the R&D activity, but also its principal input. A wide range of empirical studies
have demonstrated the positive impact of R&D spillovers on firm productivity
1
by introducing
various external inputs into the production function or cost function. These results have been
confirmed by more recent estimations (Sena, 2004, OMahony and Vicchi 2009, Autant-Bernard et
al. 2009). The empirical literature therefore emphasises the relevance of focusing on knowledge
flows and provides evidence consistent with the hypothesis of positive knowledge externalities. The
spatial dimension stressed in some of these studies is of primary interest to regional policy makers.
Following the conclusion of Marshall (1920), spatial proximity is believed to ease firm
interactions and knowledge diffusion. More recently, the Economic Geography and Endogenous
Growth models (Lucas 1988, Martin and Ottaviano 1999, Baldwin and Martin 2005) explain the
differences in regional growth paths through geographically constrained knowledge externalities at
the root of increasing returns and localised economic growth. The spatial dimension of knowledge
externalities finds empirical support with the literature referred as the Geography of Innovation ,
2

based on the pioneering works of Jaffe (1989), Acs et al. (1991), Jaffe et al. (1993), Feldman (1994),
Audretsch and Feldman (1996, 1999), Acs et al. (1997), and Almeida and Kogut (1999).
Within the literature, it is generally argued that the existence of local knowledge flows would
favour innovative capacity and drive regional economic dynamics. At the same time, however,
public policies designed to establish a technology area similar to Silicon Valley have had only
limited success and are now widely considered to be utopian (Maggioni, 2002, Rallet and Torre
2007). Thus, although the theoretical and empirical literature confirms that regional innovation
policies are important (Tdtling and Trippl, 2005; Howells, 2005; Laranjaa et al. 2008) and that they
intervene at a critical level to support institutions that generate knowledge and learning (Morgan,
2004), the fundamental question remains of how to concretely affect knowledge flows and their
geographic dimension. The answer is not obvious and is at the heart of this article.
Our analysis builds on the empirical results of the Geography of Innovation concerning localised
knowledge flows to highlight some policy implications within the European regions that support the
Lisbon Strategy. This econometric literature is interesting for regional science and innovation
policies because the results of applied econometrics provide comparisons between various
institutional contexts. Furthermore, the observation levels are generally government-defined areas,
3

which are highly relevant for regional policy makers.
Taking into account the main initial contributions founded on American and European data, this
paper is more specifically based in the latest analyses, many of which result from the European
program devoted to Intangible Assets and Regional Economic Growth (IAREG).
Recently, measurement of the geographic dimension of knowledge externalities has been
significantly improved by the introduction of more complex processes of knowledge diffusion and
the use of individual data along with new spatial econometric tools. This refines the way to model
externalities in knowledge production functions and enables testing for spatial auto-correlation (Acs
et al. 2002; Fingleton and Lopez-Baso 2006; Autant-Bernard et al. 2008).
The geographic dimension of knowledge externalities - whether they flow from science to
industry, remain intra-industrial, or are intra-firm - is confirmed by the empirical literature. This

1
See Sena (2004) for a review.
2
The term is taken from the title of Feldman (1994) which stands as a major reference in this empirical literature.
3
States or metropolitan areas in the United States, regions or departments in France.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
3

result places importance upon interventions at the regional level, which are adequate to exploit the
geographic externalities. However, because proximity is not a sufficient condition, specific actions
must be conducted to favour knowledge flows within regions. These are discussed in Section 2. In
addition, the empirical studies also clearly show that, even where they exist, the effects of proximity
are never exclusive and interact with national and international effects. Regional innovation policy
therefore has another equally important role to the promotion of local spillovers: promoting an
opening up to the rest of the world. As shown in Section 3, this involves developing the regional
absorptive capacity and connecting the regional innovation system to the national and supranational
systems, hence the explicit choice to articulate the different territorial levels of public intervention
within the European Union: regional, national, and EU-based. Subsequently, one major finding of
the empirical literature is the complexity of knowledge diffusion. The flows vary according to the
context, notably with regard to sector-based specificities, local industrial specialisation, institutional
structures, and the means of communication. Some debated questions remain concerning the
conditions under which the effects of proximity may act positively. In Section 4, we argue that the
relevant answers at the basis of any policy recommendation cannot be generalised and require case-
by-case analysis to specify the local context and strategy. Finally, a summary and concluding
comments are provided in Section 5.



2. Spatial distance matters: to what extent?


Empirical studies, including recent analyses, highlight the influence of spatial proximity on
knowledge flows and interpersonal relationships, noting that geographic proximity is a prerequisite
for learning and innovation (Autant-Bernard and Massard 2009).
However, the spatial range of the positive effects of knowledge flows is observed to vary
substantially depending on the institutional context and on the technological field (Peri, 2005).
Therefore, the existence of organisational frontiers may prevent knowledge flows. It is obvious that
such frontiers exist between public research and private firms, but they also occur between firms
(Autant-Bernard et al., 2009). The variety of knowledge may also explain its unequal transmission
through space. For example, depending upon the nature of the specialisation and upon the maturity
of the industry, the sharing of embodied knowledge may hinder knowledge flows. The diffusion of
tacit knowledge and its absorption would necessitate more than mere physical proximity. It would
rely on effective interpersonal interactions among a variety of actors. Thus, to have a significant
impact, geographic proximity effects must be combined with other dimensions such as
organisational, institutional or cognitive proximity. One major conclusion of the empirical studies
that consider the various aspects of proximity is that the geographic proximity per se is neither a
necessary nor a sufficient condition for learning. Nevertheless, it generally facilitates interactive
learning by strengthening the other dimensions of proximity and thereby contributes to the positive
effects of local knowledge externalities (Boschma 2005).
Recent advances in both the mechanisms of science-industry knowledge and in cross-industry
knowledge flows provide us with new evidence to be used in the design of regional innovation
policies. They are presented in the next paragraphs.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
4



2.1. The conditions for geographic knowledge flows


2.1.1. Spillovers from science to industry

Many papers highlight the role of public research in the production of knowledge externalities. In
most advanced countries, universities are a major source of new knowledge and a critical component
of the public research system. The public/private relationship generally refers to the relations
between universities and industry.
Following Jaffes work (1989), early studies mainly focus on the United States and are based on
data from the 1980s. All of these studies conclude that technological externalities have a localised
aspect or that the science-industry relationship is sensitive to geographic proximity within states and
metropolitan areas.
4

These initial studies were extended using data from more recent period and other countries,
providing a range of findings.
5


The role of the institutional context

Concerning France, Autant-Bernard (2001) and Autant-Bernard et al. (2005) show that
technological externalities from public research are present within innovation and private research.
The local dimension to these externalities is not very pronounced, however. One explanation for this
difference from the previous results involves the institutional factor. Despite a noticeable evolution,
links between universities and industry are less pronounced in the French institutional context than in
the US. In particular, the self-reinforcement effect between public and private research observed in
the US does not exist in France. A Paris-provinces structure is also prevalent, which could explain
weaker diffusion of knowledge.
Administrative frontiers are also likely to reduce the diffusion of knowledge, giving rise to
border effects. The intensity of knowledge flows decreases significantly beyond the countrys border,
even when regions are contiguous. Lundquist and Trippls (2009) study clearly highlights that
knowledge flows can be impeded by administrative borders, but also that passing these borders can
favour knowledge flows. Such border effects apply equally within countries due to administrative
borders between regions.
Thus, the specificity of each local institutional context and the existence of border effects
impacting knowledge flows have been highlighted through statistical analysis.






4
Mansfield (1995a); Mansfield and Lee (1996); Acs et al. (1991); Feldman (1994); Zucker et al. (1998); Acs et al. (1997).
5
Antonelli (1994), Paci and Usai (2000), Bottazzi and Peri (2001) for Italy; Kenney and Florida (1994) for Japan; Blind and Grupp
(1999), Beise and Stahl (1999) for Germany, Autant-Bernard (2000) for France.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
5


The nature of the knowledge transmitted and the temporal dimension

Other obstacles to the diffusion of knowledge over space seem to emerge from the nature of the
transmitted knowledge. Early works within the Geography of Innovation tackled this question
through the categorisation of knowledge into codified and tacit knowledge. Such a dichotomist and
static approach has since been abandoned for diverse reasons.
First, from a practical point of view, it is difficult to identify knowledge as codified or tacit, and
the correspondence between these categories and other relevant distinctions for the Economics of
Innovation, such as the public/private or academic/applied distinctions, is unclear. Second, studies
that attempt to distinguish different sorts of knowledge to evaluate their potential for spatial diffusion
have often been led to consider the temporal dimension of the diffusion process as well. It is now
clear that the spatial dimension of knowledge flows is also related to transmission over time, and not
merely to the nature of the knowledge transmitted.
Thus, to differentiate between academic and applied research, Feldman (1994) considers the
observable differences between the upstream phases of the innovation processes (research phase) and
the downstream phases (innovation, filing a patent). Similarly, for the French case, the geographic
dimension of technology spillover varies according to how their effect on R&D (upstream phase) or
patents (downstream phase) is considered (Autant-Bernard, 2001). Both studies confirm the idea of a
stronger localisation at a time when knowledge, still poorly codified, requires a physical proximity
between individuals to be transmitted. Conversely, when defined as a patent, knowledge becomes
less tacit and can be transmitted across a distance. It may also be argued that the stronger influence
of geographic proximity on research spillover than in France could be explained by the fact that
public research in the US is more "applied."

The influence of the temporal dimension on the geographic feature of knowledge flows has also
been emphasised by studies based on patent citations. Jaffe et al. (1993) for the United States case
and Maurseth and Verspagen (1999) for Europe note a decline in the localisation of patent citations
over time.
Nevertheless, this result poses a few problems of interpretation because it supposes that tacit
knowledge is more characteristic of the upstream research phases, whereas the application of
innovations would bring codified knowledge into play. No simple relationship can be established
pertaining to this point because all research and innovation activities often combine both forms of
knowledge.
Here again, the institutional context appears to be a determining factor. In particular, within the
French, and more broadly, the "European model" of public-private relationships, the transfer of
knowledge between academic research and industry occurs less via direct - hence specific and
localised - relationships than via a widespread diffusion of scientific achievements, which are likely
to benefit the system as a whole. Such an institutional organisation acts as an incentive to rapidly
codify academic knowledge (to achieve its widespread diffusion). However, at the same time, this
codification is generally insufficient for the commercial exploitation of a scientific achievement. Part
of the new knowledge remains tacit, i.e., embodied in the researchers.
This is the reason why the importance of having different diffusion channels is emphasised by
the empirical literature on spatial knowledge spillovers.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
6


The importance of the diffusion channels: effective connections and open science

A recent set of studies has explored the role played by different types of knowledge diffusion
channels between public and private research. Aharonson et al. (2007) highlight the key role of
universities in the location of new insiders in the biotechnologies industry, and the importance of
face-to-face relationships in the knowledge flows between academic researchers and firms.
As pointed out by Cohen et al. (2002) for the US and by Bekkers and Bodas Freitas (2008) for
the Netherlands, knowledge transfer channels between universities and industry are numerous,
including scientific publications, influx of students, participation in conferences, collaborative and
contract research, patents, and so on. These authors also show that the role of knowledge transfer
channels differs according to the industrial context of firms (in regards to sectoral effects and size
effects in particular).
These two studies point to the idea that the most important channels are scientific publications,
scientific collaborations and informal relationships. Moreover, they suggest that formal
collaborations are not the main mechanism for science-industry transfers. Cohen et al. (2002)
observe that the most important channels for accessing public research are the public and informal
channels (such as publications, conferences, and informal interactions), rather than formal
interactions like licenses or cooperative ventures.
In this sense, they confirm a result previously stressed by Zucker et al. (1994) concerning the
positive role played by effective collaborations. The few works that have since emerged dealing
simultaneously with direct relationships between researchers and the local dimension of knowledge
flows may be grouped into three categories.
Using joint publication data as an indicator of effective scientific collaboration, an early method
maps the coincidences between the structure of scientific relationships and the geographic structure:
see Katz (1994) on joint publications and more generally the bibliometric literature establishing
cartographies for scientific relationships.
With a more interpretative content, a second type of work ties innovation output to scientific
connections at different geographic levels. Within this type, Zucker et al. (1998) show the
importance of local relationships between companies and university researchers in harnessing
externalities. More recently, they demonstrate the so-called virtuous circle, i.e., that both academia
and firms can benefit from such an effective relationship (Zucker and Darby, 2007).
Lastly, the third type of analysis relies on the same data, but reverses the problematic using
econometric methods to assess the determinants of the observed scientific relationships. To be
specific, the aim is to define the importance of the explanations based on geographic proximity
(Maurseth and Verspagen, 1999).
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
7

All of these study types analyse the links between the geographic dimension, interpersonal
interactions and knowledge externalities. The tested hypothesis is as follows: externalities are
mediated by interactions between individuals, and these interactions are in turn facilitated by
geographic proximity. Hence, the marked influence of interpersonal relationships on knowledge
flows is undoubtedly one of the main reasons for proximity effects. Because mobility is
geographically circumscribed and the likelihood of encounters is facilitated by proximity,
publications are particularly fostered within built-up areas. However, empirical works in this field
are still rare. They come up against the difficulty of finding data that are representative of the
diversity of the formal and informal interactions that could lead to knowledge publication.
It should not be forgotten, however, that collaboration is only one strategy among others to
benefit from public research spillovers. As argued by Bekkers and Bodas Freitas (2008), if such
collaborations are efficient channels of knowledge transfer in some cases, open science channels
should not be neglected.
Focusing on these public channels, Cowan and Zinovyeva (2009) compare the role played by
scientific publications and patents as drivers of knowledge flows between public and private
research. They conclude that publications are more important than patents. The tradition of
fundamental research in the academic sphere, as well as the existence of pure knowledge spillovers
is therefore supported by these empirical studies.


2.1.2. Inter- and intra-firm spillovers

Whereas the initial studies in the field of the Geography of Innovation mainly focused on science-
industry externalities, recent advances emphasise the role of spatial externalities among firms as well
as within multi-plant firms.

Inter-firm spillovers
Recent empirical studies confirm that universities are not the only emitters of externalities and
that inter- and intra-firm localised knowledge flows have a significant impact on externalities as
well. Along these lines, Autant-Bernard and LeSage (2010) evaluate the respective roles in regional
innovation played by public and private R&D. Using panel data concerning French patents granted
to private and public research over the period 1992-2000, they estimate a knowledge production
function. The results highlight geographically bounded spillovers from public research, but also
show that the largest direct and indirect effects are associated with private R&D activity that spills
across industry boundaries. The importance of spatial distance is also stressed by Maggioni et al.
(2007) in their estimation of a spatial knowledge production function from aggregated data at a
Nuts2 level. Their results show that spatial autocorrelation prevails over relational autocorrelation,
thus supporting the idea that the regional propensity to patent benefits more from local knowledge
spillovers than from those resulting from distant collaborations.
Bergman and Usai (2009) survey the empirical studies that use patent citations as proxies of
knowledge flows, from Jaffe et al. (1993)s contribution on data provided by the United States Patent
Office (USPTO) to more recent studies on European data, principally based on data provided by the
European Patent Office (Le Sage et al., 2007; Paci and Usai 2009), or combining American and
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
8

European data (Lukatch and Plasmans 2003; Criscuolo and Verspagen 2006). They find that this
literature highlights the importance of geography on knowledge flows between patent-inventors.
This result is confirmed by Bergman and Usai (2009), who provide an empirical analysis of
patent applications and citations in the OECD/EPO and REGPAT databases to analyse knowledge
diffusion across European regions. In a first step, a descriptive analysis enables them to consider
three dimensions simultaneously: temporal (over the period 1980-2000), spatial (278 regions, most at
the NUTS2 level, in 22 countries) and industrial (22 ISIC sectors). The results show that despite a
gradual evolution, citations and patents are spatially concentrated. Clusters of innovative regions
appear, and this heterogeneity among regional flows can be related to geographic, institutional and
industrial settings. On this basis, a spatial econometric analysis is performed, based on a gravity
model, to assess the role of several potential determinants of the flows. The analysis of the impact of
space includes the role played by the removal of the barrier between Eastern and Western Europe.
The econometric results reveal the impact of geographic distance and of the weights of the origin and
destinations regions on knowledge flows, measured by GDP per capita and R&D investments. These
results vary with sector specificities.
Recent advances expand the intangible assets taken into account as a source of knowledge flows
among firms. This confirms the existence of spatial spillovers. In this manner, Dettori et al. (2009)
highlight geographic spillovers from technological, human and social capital.

Intra-firm spillovers

One main contribution of the recent studies is to provide evidence that intra-firm knowledge flows
have also an impact on firm performance and on the geography of innovation due to the location of
the plants. This original perspective takes into account the influence of intra-firm flows on
innovation and productivity.
Using French data over the period 2000-2002, Autant-Bernard et al. (2009) investigate the role of
agglomeration forces and knowledge externalities on firm productivity. They first measure
productivity changes over the period. Several components of productivity are distinguished using the
Luenberger Productivity Indicator. A Maximum Likelihood Random Effect model is estimated in a
second step to study the influence of intangible assets on productivity.
The most interesting result provided by this study for a knowledge-based regional policy is the
impact of intra-firm spillovers across multiple locations whose influence on the technological
component of productivity is even stronger than that of inter-firm spillovers. This result draws
attention to the importance of knowledge management.
This is precisely the focus of Schienstock (2009). As with the preceding contribution, this paper
shows that in addition to the need for firms to acquire external knowledge, internal knowledge flows
must not be neglected. From a firm survey conducted on a sample of medium- and low-tech Finnish
firms, this study investigates the impact of organisational and knowledge management methods on
firm performance. Several objective and subjective performance indicators are taken into account.
The results emphasise the positive impact of knowledge management on firm performance. More
specifically, the study evaluates performance measured in terms of innovation capacity and
efficiency of collective learning. In addition, the study shows that there is no generic best practice in
knowledge management. This means that the best strategy in this field depends on the sector and
more generally on the context, as well as that the strategy has to be firm-specific.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
9

2.2. Policy implications: fostering local knowledge flows


The local economic system is unique, and the existence of geographic externalities bears witness
to the idea that regions are characterised by certain non-transferable interdependencies. Being neither
static nor irreversible, such specific interdependencies are subordinated to the public action.
Different combinations of proximity can be relevant in different local contexts.
Hence, a challenge for local policies is to contribute to the self-sustainment or reinforcement of
the local knowledge-based dynamics within a changing context. This is why regional factors are key
in innovation policies. Bearing in mind the principle of subsidiarity, the regional level appears to be
adequate to exploit the diversity of the local technological connections and to valorise regional
comparative advantages in technology. From the empirical literature presented above, which stresses
different contextual elements likely to hinder knowledge flows at the local level, we draw some
policy orientations. They involve bringing together different spheres that do not "naturally" exchange
knowledge: science and industry, different types of industries, and different types of firms.


2.2.1. Relying on the presence of a university


Ensuring the compatibility of two logics

The results of the empirical literature devoted to geographic knowledge spillovers show that,
within some institutional contexts, public intervention is required to improve the links between
science and industry at the regional level. This finding refers to the well-known debate concerning
The European paradox, according to which Europe is characterised by excellence in research
together with a limited industrial base in terms of innovation, in addition to the European delay in
relation to the US regarding the transfer of technology and knowledge between universities and
industry (European Commission, ERIK, 2007). Our perspective however, differs strongly from a
simple catch-up approach and, in contrast, insists on the importance of the institutional and
geographic context that underlies knowledge flows.
Despite its strategic importance in a "science-based economy", the science-industry connection
may turn out to be inadequate because it is not natural. Universities, which occupy a central place in
the generation of knowledge, and industry belong to two different worlds with specific codes,
cultures, reward systems and final objectives. Such characteristics involve bridging problems
between these two spheres.
A second fundamental reason legitimising public intervention in this field is the need to monitor
the conditions under which science and industry unite. Indeed, this is accompanied by the danger of
nullifying the advantages of "open science" (Dasgupta and David, 1994).
6
The scientific community
has traditionally played a key role, not only in the creation of knowledge but also in its widespread
diffusion. In this system, the "knowledge dilemma
7
is resolved by means of remuneration specific

6
This is the pioneering work on the opposition between private technology and open science.
7
The dilemma between the logic of opening-up - that is to say the widespread diffusion of knowledge that increases its social value -
and the logic of clamping down (such as intellectual property rights) necessary to stimulate research (Fadairo, 2002).
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
10

to the university (reputed within the scientific community through publications), which ensures an
effective compromise by simultaneously stimulating research and knowledge communication
initiatives.
Today, the knowledge dilemma is stronger whenever there is an evident link between the
scientific and industrial domains (Joly, 1992). Academic research, which generates strong knowledge
externalities, has traditionally been conducted in the public domain. For this reason, the nature of the
technological knowledge as public property (Arrow, 1962) did not pose a problem. In addition, the
patent has traditionally been exclusively used by industry. Founded on exclusion,
8
this incentive
system has shown itself to be adequate, bearing in mind the fact that the externalities of the applied
research are weak.
We are witnessing a growing tendency to protect the knowledge resulting from public research (a
consequence of a new concern to valorise the results of this research), as well as the development of
externalities resulting from private research (a consequence of the increase in private funding of
R&D activities and the growing involvement of large industrial groups in basic research). This is the
reason that universities are decreasingly the only generators of new knowledge, although they are
more often the central point of the networks of public actors and private participants in knowledge
generation and diffusion. Foray and Mairesse (2002) argue that such specificity constitutes the very
definition of a knowledge-based economy: an economy in which knowledge externalities are more
powerful than before. This does not change the nature of the knowledge dilemma, only its intensity.
This is why "the institutional compatibility of open knowledge with private incentive structures is
one of the most important compatibilities for the future of knowledge-based economies." (Foray and
Mairesse, 2001).
It now seems vital to emphasise the interaction between the two spheres of knowledge
represented by universities and industry. Universities are under pressure to open up to external
players (particularly firms) and to discover how their research results can be valorised to contribute
to regional economic development. However, to realise the benefits of the interaction between
science and industry, each sphere must retain its own specificities. Indeed, according to Dosi et al.
(2006), the European Paradox would not be global, European paradoxes would appear at the two
extremes, with weaknesses in both the scientific research systems and the corporate industries. This
explains why interactions between science and industry should not be fostered as a negation of the
specific role of each sphere, but instead to reinforce the role of each. Among other things, this means
that universities should not be confined to the role of a service provider for industry.
A related issue pertains to the ability of the ERA to retain qualified academic personnel. The
results of Bergmans survey (Bergman, 2009) show that the main reasons underlying the preferences
of academic personnel for destinations outside the EU are the lack of attractive salaries and research
opportunities offered within the ERA. Therefore, policy makers can foster attraction to the EU by
improving these two factors.






8
Because it establishes a monopoly of exploitation.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
11

Promoting science-industry transfers and cooperation

The objective is to enhance the transformation of scientific results into competitive performance,
or, in other words, to improve the diffusion of academic knowledge throughout local industrial
structures, while ensuring that the "open science convention" (Foray, 1997) is not fundamentally
challenged.
In spite of the main differences between the scientific and the industrial fields outlined above, it
is possible for these two areas to interact due to the existence of common or complementary
objectives on which public intervention can be based. Hence, the accumulation of knowledge has a
common objective for both the industrial and the scientific fields. Moreover, there are emerging
complementarities between the search for technological advances in industry and the search for
financial resources at the scientific level.
At stake for public authorities is the implementation of "distribution-orientated institutions" that
favour the diffusion of technological knowledge (David and Foray, 1994) while ensuring that the
level of research incentive is sufficiently high.
The science-industry relationship can take diverse and complementary forms according to the
extent of its embodiment: it can be anything from a simple transfer to complete cooperation between
the two spheres.
At a first glance, it may seem that regional authorities have little to do with the disembodied
channels. As argued by Fadairo (2002), because of its informational content, the patent is the most
disembodied transfer medium between science and industry, along with publications specific to
"open science". For this reason, this institutional mechanism (industrial propriety) is at its most
effective level when it operates at the highest territorial scale - at the level of the European Union.
However, the regional authorities have a double role to play here: both facilitating access for local
firms to the information contained in the patents, information, advice, and tax incentives for license
purchasing; and encouraging local universities, which in the European case are not accustomed to
this approach, to adopt patenting as a diffusion channel.
Nevertheless, the interaction of science and industry at regional level more often takes a more
embodied form, whether formal or informal. Hence, a number of studies that examine the local
character of relationships between universities and industry note the importance of informal links
like seminars, consultations or visits to laboratories. These informal links are a good way to provide
opportunities for interaction.
Formal relationships between science and industry take the form of contractual arrangements
with varying durations, including university funding granted by private companies, research shared
between public and industrial laboratories, and the "hiring" of students.
The highest degree of transfer embodiment is the creation of incubators by university laboratories
("academic incubators") that accommodate and support the project carriers prior to the formation of a
company. Analysing the impact of labour mobility, Audretsch and Keilbach (2004) and Schiller and
Revilla-Diez (2009) recognise the spinoffs and startups initiated by university scientists as key
drivers of knowledge flows. Almeida and Kogut (1999) and Breschi and Lissoni (2003, 2006)
indicate that the higher the mobility of scientists, the greater the diffusion of knowledge. This labour
mobility not only refers to spatial mobility and career mobility (mobility from one position to
another), but also to organisational mobility from public research to industry. The science-industry
relationship may also take the form of cooperative/joint research, which goes far beyond a simple
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
12

transfer. In this case, integration exists, for example, through the formation of the common institution
of a joint research centre. Such a sharing of research between the academic sphere and the private
sector involves the joint definition of output, a long-term strategy, and a mutual acquaintance with
working habits.
It therefore appears important that the regional technology policy provides the incentives
necessary for the development of a variety of forms of transfer and cooperation between science and
industry.
9
This assumes the emergence in the academic world of an open attitude towards the local
economic environment that favours the valorisation of the research output, as well as the supply of
advice or training to companies. This attitude must not hide the essential mission of the university,
which is to create knowledge and to ensure that it is distributed widely through publications and
training.
Finally, whatever modalities of interaction are established between universities and industry, the
local influence of the university still actively depends on the environment in which it is located.
Depending on whether it is surrounded by high-tech industries or mature industries, more specialised
or more diversified industrial structures, large companies or micro-businesses, a university will
develop different potentialities. In addition, as showed by DEste and Iammarino (2009), the features
of the university itself influence the manner of local cooperation. Analysing the specific nature of the
relationship between firms and universities through joint research, DEste and Iammarino clarify the
spatial profile of relations that require face-to-face contacts. Based on the UK case, they show that
differences exist in the spatial patterns of cooperation, depending on the sector and scientific fields
concerned. Moreover, the quality of academic research - measured according to a ranking of UK
universities - also has an impact on the spatial dimension of cooperation: large and top-ranking
universities display a higher propensity to collaborate locally.
To favour science-industry transfers and cooperation, regional policies must account for all of
these local features. Region-specific studies are therefore necessary to better understand local
characteristics.


2.2.2. Local networking and knowledge management


Bringing different spheres together

Measures that encourage direct inter-industrial contacts stimulate an exchange of the tacit
knowledge accumulated within firms. This type of interaction is important because it restricts the
phenomenon of the depreciation of pertinent technological knowledge over time. This public policy
objective attempts to create a situation similar to Antonellis (2002) observation that within districts,
knowledge assumes the characteristics of a collective activity resulting from the common effort of a
variety of connected agents. Even if one admits that it is deceptive to try to artificially reproduce this
type of spatial organisation for innovation activity, the reality of technology districts in this case
provides lessons that are likely to steer public action. Moreover, it should be noted that accentuating

9
It is interesting here to refer to the emerging empirical literature on the role and efficiency of Technology Transfer Offices (Conti et
al. 2007)
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
13

the collective character of innovation may attenuate the public/private dilemma surrounding
technology knowledge.
Such policies are designed to encourage companies from different fields to collaborate. In
particular, by promoting linkages between high-tech sectors and traditional industries, public
authorities can improve the diffusion of generic technologies and the hybridisations that are sources
of innovation.
Stimulating the generation of variety is also a major role assigned to technology policy by
Metcalfe (1994) and Cohendet and Llerena (1997). Expanding diversity means increasing the number
of available technical options. In short, the role of diversity within innovation is of particular
importance in the current era, when innovation occurs mainly through recombination. Empirical
research has shown that this diversity may be important for the more traditional sectors, which find
their innovation opportunities externally.
Experience has shown the difficulties in the implementation of local multidisciplinary
cooperative structures generating collective innovation processes. There exists a problem of how to
combine these institutional networks, which are created and maintained by regional institutions, with
the local industrial system. Further, setting up transverse cooperation between the local players
becomes difficult when these players do not enjoy organisational proximity.
Because of this, cooperation cannot be imposed. Intercompany contacts, however, may be
stimulated by local programs that use a good understanding of local actors to support cooperation
projects. In this context, it seems important to encourage trans-sector based cooperative structures
and, on a wider scale, meetings on a transverse theme. In addition, numerous theoretical and
empirical papers highlight the importance of supply-demand relationships in providing new
opportunities for inter-sector cooperation. It may therefore be advantageous to move forward with
policies that are purely oriented towards research and innovation supply to promote local demand.
Studying the extent to which the innovative performance of firms depends on the type of
cooperation they develop and how they reveal regional differences, Iammarino et al. (2009) have
demonstrated the positive effect of diverse types of cooperation: between firms, with suppliers, with
consumers, with public research laboratories, and with institutions of higher education.

Access to knowledge management within local businesses

As mentioned earlier, Autant-Bernard et al. (2009) and Schienstock (2009) provide evidence that
intra-firm knowledge flows must not be neglected because of their positive impact on individual
performance (in terms of productivity or innovation). Moreover, Schienstock (2009) highlights the
potential of knowledge management methods and organisational innovation.
These results define a field of intervention for the regional policy. They demonstrate the benefits
of helping local SMEs to integrate recent methods of knowledge management. This can be done by
organising seminars for local firms; the objective here is to develop awareness. In addition, it seems
necessary to provide SMEs with specialised technical consultants to help them to design methods
appropriate to their particular case.
Indeed, Schienstock (2009) emphasises the importance of flexibility because there is no generic
knowledge management model. Such a result reflects the specificity of each enterprise, sector-based
specificities, and the impact of the changing environment, which mitigates the relevance of
generalisations.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
14


3. Local/Global connection and absorptive capacity

The empirical literature shows that when they occur, knowledge spillovers are not purely local
phenomena. They appear to be simultaneously local and global and to emanate from a variety of
sources. In fact, when supported at various geographic levels (for example, counties, metropolitan
districts or states in the US), the existence of externalities internal to the zone is revealed each time.
When studies compare different geographic scales (for example, Autant-Bernard 2001, for France
and Bottazzi and Peri, 2003, for Italy), several levels of diffusion appear, even if the local effects
take precedence in certain circumstances. Such combinations are related to the manner and the
conditions in which technological externalities are transmitted, which partly refer to the concept of
absorptive capacity (Cohen and Levinthal, 1989). In the next section, new evidence on this issue is
reviewed and implications for regional innovation policy are drawn.


3.1. Shortcomings in the regional absorptive capacity


3.1.1. Asymmetries regarding the level and diversity of internal competencies

It is now widely acknowledged that an initial condition necessary for the flow of knowledge is
the constitution of an absorptive capacity. According to Dosi (1988) and Cohen and Levinthal
(1989), internal skills and competencies (internal research, diversity of available competencies) are a
precondition to capture technological externalities, or to identify and exploit knowledge that is
newly available in the environment. This idea, however, has stimulated few empirical works at the
aggregate level of geographic zones.
From their analysis of the French case, Autant-Bernard and LeSage (2010) argue that the
research level and its degree of diversity may not simply affect the level of externalities captured
within a local context, but also their geographic origin. A high and varied level of internal
competencies appears to be vital in the capacity to take advantage of remote knowledge sources.
Conversely, zones that are not very active on the research side or that are very specialised seem
better able to take advantage of neighbouring sources of externalities. Therefore, the absorptive
capacity relies more on the capacity to tap into remote sources of externalities than on the level of
externalities captured.
Paci and Batteta (2003) find that differences exist between traditional and more innovative
industries. Geographic distance reduces the technological flows in traditional sectors that are
characterised by tacit and un-codified knowledge. In high-tech sectors, the exchange of codified and
standardised knowledge can take place even among firms located in remote regions.






h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
15

3.1.2. Regional lock-in

Asymmetries of firms or geographic zones in their capacity to absorb external resources appear
to be a major problem when added to the localised feature of technology (Antonelli, 1999; Metcalfe,
1994). Path dependency in technology explains the possibility of a regional lock-in, which means
that the local tight industrial network excludes strategic knowledge (Carlsson and Jacobsson, 1997;
Ernberg and Jacobsson, 1997). This risk is also related to local institutions, which may stimulate or
hinder innovation depending on the context. The risk is even greater if we consider that institutional
change is characterised by a phenomenon of inertia, which generally makes change incremental and
slow. Therefore, regional dynamics may be positive, but they also involve a risk of technological
lock-in. A firm may be stuck with an old technique because the local system does not supply the
appropriate technology.
For this reason, public intervention is justified so that regional dynamics becomes an asset for
local firms instead of a weakness. The objective here is to provide the conditions necessary for the
evolution of the regional innovation system. This objective requires the systematic encouragement of
opening towards the outside and diversity in the broadest sense. This point is crucial in a time of
change, which triggers a greater level of uncertainty (Johnson, 1992).


3.1.3. Barriers to entering networks

One of the main features of the knowledge-based economy is the importance of international
networks through which pertinent knowledge flows (Lundvall, 1998, David and Foray, 2002).
Strategic skills are developed in an interactive manner and shared within networks. The empirical
literature on knowledge spillovers takes the impact of scientific and technological networks into
account. These international networks are likely to create social forms of proximity that favour the
transfer of knowledge. The intensity of bilateral collaboration determines the relational proximity
between agents. As stated by Amin and Cohendet (2004), relational proximity can compensate for
the absence of spatial proximity and enable long distance knowledge flows.
Recent advances on this issue come from two different research fields. The first, stemming
directly from the Geography of Innovation, involves social network analysis (Granovetter, 1985),
whereas the second comes from the theory of network formation (Jackson and Wolinski, 1996),
which emphasises the role of network effects (Jackson and Wolinski, 1996, Bala and Goyal 2000).
To assess the benefits of a bilateral relationship, we must take into account not only individual
knowledge partners, but also the knowledge that each partner can access through his/her own
collaboration networks. Therefore, the agent position within the network is considered to be a key
determinant of the ability to benefit from knowledge flows (Bala and Goyal, 2000)
Studies based on patent data highlight the relevance of social proximity in evaluating the degree
of knowledge spillovers: they consider the influence of social distance on the probability of citing a
patent (Singh, 2005, Sorenson et al., 2006, Gomes-Casseres et al., 2006).
Varga and Parag (2009) analyse the structural characteristics of networks as underlying
mechanisms of knowledge flows, as well as the impact of individual positioning within networks.
They show that the capacity of the research centres of the University of Pecs to file patents is linked
to their position within the international network of publication. Several characteristics of the
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
16

network structure, such as its concentration, size and the degree of integration are included as
explanatory variables within a knowledge production function. The higher the quality of the network
positioning, the greater the number of patents filed.
Finally, as noted by Autant-Bernard and Massard (2009),
10
a mix of connectivity based on
geographic proximity with emote connections is often considered the best way for a network to
induce efficient knowledge flows. Access to these international networks, however, is not open and
free. Among other things, it presupposes the sharing of tacit knowledge or of codified knowledge
where the codes are difficult to acquire (Cowan and Foray, 1997; Maskell, 2002). However, the
capacity to join these tight networks determines the access to knowledge, which is currently the most
strategic resource. The existence of barriers to entering networks in which knowledge is produced
and transmitted presents a field of intervention for regional innovation policies.



3.2. Policy implications: connecting the local innovation system to international levels


3.2.1. Learning

The concept of "a learning economy" (Lundvall, 1998) synthesises the idea that if knowledge is
currently the most strategic resource, then learning constitutes the most important process in
economics. Accessing scientific and technological knowledge does not simply presuppose that the
system has the "distribution power" (David and Foray, 1994) necessary to ensure availability of this
input, but also that firms have the capacity to absorb external resources. This is a very challenging
exercise given the current speed of development. From this perspective, it is clear that there is a
minimum activity threshold in research below which "nothing happens", which means that no
learning dynamic is likely to develop. This involves a specific challenge for small structures - firms
or regions.
If the available diversity of skills is to be favourable for local learning, the development of poles
of specialisation must not be overlooked, because without them, it seems impossible to capture the
leading knowledge from distant centres of excellence (see Autant-Bernard and Massard, 2009).
Because learning is deeply affected by institutional architecture (Carlsson and Jacobson, 1997),
public authorities are responsible for developing the means of learning and the capacity to
communicate. Such a target presupposes long term interventionism because learning is far more than
a transfer of information and cannot be reduced to a single transaction (Fadairo and Massard, 2000).
For this reason, we argue that an education policy is an integral part of the innovation policy, which
extends beyond the quantitative issue of funding. The education system is involved at every level:
"from nurseries to the training of engineers and scientists" (Lundvall 1992).



10
See also Autant-Bernard and Massard (2009) for a survey concerning the applications of game theory within the theory of network
formation.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
17

More concretely, developing an education system requires improving physical infrastructure,
equipment and human resources. Here, the key role played by universities becomes apparent
(Caracostas and Soete, 1997; Lundvall, 1992). Training also helps maintain the "creative forgetting"
that is necessary to move from one technology to another. In addition to developing the means to
learn, public interventions may also help stimulate a culture to learn. As an example, one such
intervention might involve individual rewards for learning and creative efforts (Lundvall, 1992).
Education may also favour better integration within international networks as well as a better
diffusion of knowledge. Guiri and Mariani (2007) observe that the level of education has an impact
on the size of the network: The higher the level of education, the larger the network. Peri (2005) also
finds broader spatial knowledge spillovers that stem from the leading regions.



3.2.2. Communication infrastructures

As mentioned earlier, the results of the literature on knowledge flows favour innovation policy in
the broadest sense, including public actions in education and training. This long-term interventionism
involves establishing conditions that are favourable for innovation instead of providing direct,
targeted intervention. This characteristic also occurs in another measure necessary for the capture of
external knowledge: the development of all forms of communication infrastructure.
It is worthwhile to make several comments on this point. First, in the ideal situation defined by
Dasgupta and David (1994), access to new knowledge is broad, fast and free. These features
determine a "system's distribution power" and naturally depend on the quality of the communication
infrastructure. From this point of view, all actions encouraging the codification of new knowledge
constitute the first stage in communications policy.
In addition, it is important to have a diversity of means of communication. This is because new
information and communication technologies are critical in determining access to external
knowledge. Promoting their development and diffusion at the local level diverts potential
partitioning. Even though the geographic dimension still has an effect as shown by econometric
studies these technologies considerably weaken the constraints of physical distance. The role of
spatial dispersion is therefore secondary to the role of professional cyber-communities, which share a
code, a language and more: a culture.


3.2.3. Access to European programs for local businesses

In the European Union, the existence of a community technology policy is an undeniable asset to
the regions. The formation of international cooperative structures, driven by the European programs,
gives local businesses the opportunity to correct any weak points in their absorptive capacity. It is a
means of escaping from the dependency on the local path to access closed networks and tap into
international technological externalities.
This participation in European programs constitutes a major element in the local dynamism of
small businesses, which largely rely on external sources of knowledge (Audretsch, 1995). Today, a
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
18

major problem is that despite efforts,
11
the proportion of SMEs participating in European programs
and their capacity to benefit from those programs remain low.
This is an important field of action for regional innovation policies: to remove institutional
barriers and encourage the participation of SMEs, and more generally of regional firms, in European
programs.
The role of training and codification discussed above, merits further emphasis: they are the
conditions necessary to access common and evolving languages. Moreover, advisory activities for
SMEs at a regional level should be encouraged to allow them to join the European collaborative
structures.


3.2.4. Attractiveness

Finally, an important aspect of local/global interactions is the ability of a region to attract
workers, researchers, professors and firms, which are important channels of embodied knowledge.
Note that the objectives presented above (learning, communication infrastructure, local businesses
access to European programs for) concern both the regional absorptive capacity (i.e., the ability to
capture diffuse external knowledge) and regional attractiveness (i.e., the ability to capture external
embodied knowledge). This second point refers to the quality of the local environment and the
promotion of a comparative regional advantage. It must be mentioned that attractiveness of a
territory depends upon its dynamic comparative advantage, which goes beyond the static competition
factors provided locally such as costs and wages.
Investment in physical capital plays a major role in creating local public goods contributing to
competitiveness (OECD, 2009). Infrastructure and quality services appear essential to attract new
skills. For example, communication and transport infrastructures play a key role for companies
because they promote access to a wider reservoir of manpower, offering a broader range of skills.
They also favour faster access to suppliers and customers.
However, it is known that quality infrastructure is not sufficient for territory attractiveness.
Focusing on the provision of communication and transport infrastructures can generate a
phenomenon of leaking by linking (Hirschman, 1958). Whether transport infrastructure becomes a
binding factor (link) or a loss factor (leak) depends on its insertion into a wider regional strategy
that is business friendly, skilled-worker friendly and that limits the brain drain.
Among the various measures that provide a supportive environment for the local economic
actors are the promotion of human capital, universities and learning. Research centres and
universities help produce highly qualified personnel and new knowledge to be used and adapted by
local firms. They therefore appear as a key factor of attractiveness.







11
Of the European Commission, notably, but also in many case of national policies.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
19

4. The need for original strategies

Finally, a major finding of the empirical literature devoted to spatial spillovers emphasises the
complexity of knowledge diffusion and the great diversity of contexts. According to Boschma
(2005), different types of proximities related to knowledge flows can emerge, develop, and
disappear; they might not only reinforce each other, but also sometimes act as substitutes. For this
reason, local knowledge externalities are not unchanging and universal. Consequently, there is a need
to consider the dynamic nature of proximity interactions and to take into account that the regions
display different capabilities to benefit from local knowledge flows.
This is why, beyond the general recommendations made above, there is room for specific and
dynamic regional strategies. Howells (2005) highlights the diversity of possible approaches to
regional innovation policy. Tdtling and Trippl (2005) stress the need for differentiated regional
innovation policies, showing that there is no ideal model because innovation activities differ greatly
between regions. It is therefore unrealistic to search for universal policy tools.
In addition, analysis of the results obtained in the literature shows numerous ambiguities that are
central to the following public policy debates that generally necessitate strategic trade-offs: equity
vs. efficiency, commercial valorisation vs. scientific excellence, and specialisation vs. diversity.
These dilemmas are discussed in the following sections.


4.1. The equity/efficiency dilemma

The equity/efficiency dilemma appears as a main result of the Economic Geography-Endogenous
Growth models, which show that the agglomeration of R&D means and innovative activities may
lead to unequal levels of economic growth across regions and to greater economic growth at the
global level. This result is based upon the hypothesis of the existence of local barriers to the diffusion
of knowledge flows. The current literature on the mechanisms of knowledge diffusion, however,
leads us to question the relevance of such a hypothesis because this literature takes also into account
the existence of remote knowledge flows. It is crucial to note that knowledge production may result
from diverse forms of knowledge diffusion using local and global linkages with variable intensity.
Some forms are still strongly based on the existence of agglomeration economies and push the
development of greater and increasingly diversified agglomerations of innovative activities. Other
forms rely on fine specialisation dynamics and generally require relationships between smaller
clusters. Hence, higher efficiency does not always correspond to higher concentration.
It is important to acknowledge the diversity of the forms of innovative concentration that can
give rise to a great variety of local strategies. Some strategies offer a variety of opportunities for
connections without being obliged to give specific incentives or to organise these exchanges. In
contrast, other strategies are based on the quality of the relationships and on the organisation of
knowledge exchanges between a reduced number of very specialised and well-known actors. The
relevant question becomes how to efficiently articulate these diverse spatial organisations of
knowledge flows.
The results obtained by Bergman and Usai (2009) highlight that, within the EU, knowledge flows
(evidenced by patent citations) are strongly and permanently localised within core countries. They
come from a small number of strongly agglomerated areas (see also Peri, 2005). In this context, the
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
20

main issue concerns how to design policy at both the national and European levels while taking into
account the concentration of the most frequently cited patents in the various centres of excellence
throughout Europe. Bergman and Usai defend the idea that selective divergence may be necessary in
the development of these critical capacities. This is why policy efforts should help attain and
reinforce innovative excellence, which can then be shared with less innovative regions, rather than
continue attempts to disperse the capacities to generate large numbers of patents. Finally, they
suggest that more attention should be paid to the importance of innovation at border regions and that
policies should be adjusted to tap these newly developed capacities to benefit regions at the interior
and in more peripheral areas.
In these peripheral regions, the main problem is one of absorptive capacities. It is necessary for
policy makers to determine the changes that are required at the local level to benefit from the
innovation developed in remote poles of innovative excellence. There is no universal response to the
question of critical economic weight. Instead, the answer will result from public local authorities
specific response to the following question: what are the minimum locally required means to engage
a virtuous circle, whether based on pure agglomeration or on fine specialisation and integration in
global networks?


4.2. Dealing with the commercial valorisation / scientific excellence dilemma

Different conceptions of efficiency may be considered. Within the Economic Geography-
Endogenous Growth models, local knowledge externalities enhance local innovative efficiency by
reducing the research cost, which allows new industrial activities to emerge. However, different
conceptions of efficiency in terms of knowledge creation can be distinguished that underlie public
policy decisions on science and technology. Some emphasise scientific breakthroughs from star
scientists who diffuse their knowledge via scientific publications, whereas others emphasise the
success of commercial applications of science. This fosters important debates because it is widely
believed that the involvement of scientists with firms comes at the expense of scientific progress. In
contrast, however, recent studies from Zucker and Darby (2007) prove that the commitment of talent,
time and energy to firms would not come at the expense of science. Indeed, investigating the
biotechnology case, Zucker and Darby (2007) reveal the existence of a truly virtuous circle of
interactions between public and private research: private research is not the only beneficiary of such
relationships because R&D cooperation connections with firms improve researchers academic
output. Within a context where the role of open science is reaffirmed, developing collaborative
activities between public and private researchers is a good way to boost the dynamics of such a
virtuous circle and to deal with this dilemma.


4.3. Dealing with the specialisation/diversity debate

A highly debated question deriving from the empirical literature on localised spillovers is
whether local inter-firm knowledge flows are more encouraged by a specialised environment or by a
diversified environment. We refer to the debate revived by Glaeser et al. (1992) that opposes
Marshall-Arrow-Romer-type externalities resulting from interactions between specialised agents
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
21

with similar competencies to Jacobs-type externalities based on the combination of diversified
activities. As showed by Autant-Bernard and Massard (2009), the empirical literature does not allow
us to decide this theoretical debate because the effects of sector diversity are heterogeneous with
respect to the area considered in terms of the types of industry present locally, the degree of
openness, and other factors. For example, De Groot et al. (2007) reveal the impact of the population
density of the region: denser areas are more likely to exhibit positive effects of specialisation. This
means that a single and simple policy rule does not apply.
From a policy point of view, Foray et al. (2009) propose the concept of Smart Specialisation,
which supposes that specialisation strategies are chosen by each region in accordance with its own
industrial characteristics and its relationships with other regions. This framework suggests strategies
that can be pursued to the advantage of both regions that are at the scientific and technological
frontier and strategies that are less advanced. Whereas the leader regions invest in the invention of a
General Purpose Technology (GPT) or the combination of different GPTs (bioinformatics, for
example), follower regions often are better advised to invest in the coinvention of applications,
that is, the development of the applications of a GPT in one or several important domains of the
regional economy. This may provide a concrete foundation for the idea of selective divergence
proposed by Bergman and Usai. Hence, the specificity of the industrial structure of each region has
to be taken into account to create an original strategy for each case.
This requires accurate knowledge of the local features and of the main strengths and weaknesses
of the area. This calls for additional analysis based on smaller geographic and sectoral levels. The
lack of localised data on innovation is therefore a major challenge from a policy point of view.


4.4. Policy implications: building specific local strategies

Knowledge of the local context and its global position is the most fundamental condition to define
an original innovation strategy at the regional level. Unfortunately, the studies engaged in the
IAREG project point to a lack of relevant data. Knowledge flows are by definition a phenomenon
that is difficult to measure. However, some indirect indicators have been suggested during the last
two decades. Patent citations, as a paper trail of knowledge flows, were the first suggested indicators,
from the seminal work of Jaffe et al. (1993). Other indicators include the use of an external R&D
pool in other industries and/or other areas, the use of intermediate inputs (based on input-output
matrices or on technology flow matrices), labour mobility between firms or from science to industry,
and integration within local and global networks.
Most of these data are available at aggregated levels. Several datasets are provided by each
country in the EU or by the OECD. Some noticeable attempts to regionalise this information have
been made recently (see, for instance, the regional patent OECD database). However, systematic
information is still missing at the small regional scale and at the disaggregated sectoral. Studies on
productivity highlight that knowledge diffusion mechanisms within and between regions favour
productivity and growth. There is an urgent need to obtain precise information about these
knowledge flows.
Therefore, the implementation of technological policies aimed at fostering knowledge flows
between EU regions requires:
- Easier access to data. At the moment, the data are dispersed among various statistical offices
(Eurostat, national statistical offices, international offices, etc.). It is therefore time consuming to
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
22

obtain the data from these different organisations and then to build homogeneous databases for
common data.
- More systematic collection of the data. Some data are collected only within certain countries or
regions. This prevents the development of a general overview at the EU level. Other data are
collected at the EU level by Eurostat. However, substantial missing observations prevent an
exhaustive picture of the EU regions. Comparisons between sectoral and spatial data are especially
difficult. Other data come from the OECD or from international reports that do not cover all EU
countries.
- Improvement and diffusion of the indicators to adapt them to local needs. This requires organising
the interactions between the data producers, policy makers, and researchers who specialise in these
topics. For example, according to Bergman and Usai (2009), the European Innovative Scoreboard
includes EPO patent application rates, which indicate quantitative levels of innovative effort but not
its quality. Adding the number of OECD country citations to patents filed in each region as a
measure of their quality may improve the attributes of the indicator. Moreover, the indicators of
knowledge flows developed in the academic sphere are not yet available for policy makers. The
information available at the local level relies on input and output indicators of innovation. This is
insufficient to position a territory and to evaluate the impact of past technological policies, which is
required to enable a learning process in public policy (Nauwelaers and Wintjes, 2008). It is crucial to
provide policy makers with a detailed accounting of innovation in EU regions, especially regarding
the local ability to benefit from local and global knowledge spillovers. At the same time, policy
makers have specific needs that must be expressed to improve the data. Fostering interactions
between the different actors would favour feedback likely to enhance our ability to understand and
benefit from knowledge flows.
This is of interest for scientific research. Indeed, several points remain uncertain regarding the
mechanisms of knowledge diffusion due to a lack of data. For instance, it is very likely that better
disaggregation of the regional R&D data by industry would allow us to give new evidence on the
MAR vs. Jacobs debate on intersectoral knowledge flows. Similarly, limited information is available
regarding labour mobility and network relationships. Improving the available data would allow us to
deal with these new issues, and help in the design of future technological policies.
This is also of direct interest for current technological policies. Policy makers, especially at the
local level, are increasingly expected to spur innovation dynamics. Increasing financial resources are
devoted to local actors (regions, agglomerations, clusters, etc). However, the available tools to build
relevant strategies and to monitor these innovation policies are still rare.
A short-term improvement of the available data would therefore rely on the construction, at the
EU level, of a specific place where data could be centralised and homogenised, where relevant
indicators could be discussed and made available, and where information could be diffused, both to
the scientific community and to policy makers.
12


12
Such a structure has been tested in France within the EuroLIO project. Its expertise in spatial data analysis and in the diffusion of
technological information could be extended to the EU by including EU actors (research labs, data producers and representatives of
policy markers). The French project implemented a Localised Innovation Observatory (EuroLIO, ww.eurolio.eu), which takes the
form of a network between five research labs specialising in localised innovation dynamics, the innovation data producer offices and
representatives of the national and regional technological policy makers. This is a specific place where data is centralised and
homogenised, where relevant indicators are discussed and made available, and information is diffused, both to the scientific
community and to policy makers.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
23


5. Summary and concluding comments

In the context of multi-territorial governance, the regional level of intervention appears to be
decisive in favouring the establishment of a regional comparative advantage based on knowledge
flows between local public and private institutions. Indeed, both theoretical and empirical studies
point out the specific role of space in knowledge production and diffusion. The most recent empirical
studies provide us with new insights on the underlying mechanisms. The role of formal and informal
interpersonal relationships and the role of mobile scientists as agents in knowledge spillovers show
that the local diffusion of knowledge is neither obvious nor systematic. Policy makers can therefore
favour knowledge diffusion between local actors by fostering scientists mobility and interpersonal
relationships to bridge different worlds (science-industry as well as across industries).
However, considering the complexity of knowledge diffusion, which is only partly localised,
and the risks of regional lock-in, regional innovation policy must also encourage the entry of local
actors in international networks where pertinent knowledge is created and exchanged. Indeed,
theoretical approaches to network structure, confirmed by empirical studies, stress the importance of
both intense local connections and global ties. However, pure networking policies, whether local or
international, are not sufficient. Networks are more efficient when each node is sufficiently strong by
itself and has a specific function within the process of knowledge production and diffusion. Because
of this, building research potential and absorptive capacity is a key factor in firms integration within
international research networks. Technological policies promoting private R&D not only favour
firms own R&D activities, but also firms ability to benefit from research network spillovers.
Public incentives for R&D collaborations appear to be a complement, and not a substitute, to public
support to private R&D.
Finally, an important outcome of recent empirical studies is the great variety of regional and
sectoral contexts. There is, in particular, substantial diversity in spatial organisation, with some
regions highly agglomerated and diversified and others relying on a lower level of agglomeration
associated with a strong industrial specialisation. This result supports the idea that polarisation in
very large cities is not the only model of spatial organisation likely to favour innovation production
and diffusion. This explains why each region must create its own dynamics.
Innovation policies may therefore devote specific attention to lagging regions. Helping lagging
areas reach a critical mass will allow them to benefit from knowledge flows within and across the
region. In particular, policy makers should focus primarily on medium-sized regions that require
initial assistance to access to global knowledge flows, instead of devoting too much attention to the
largest areas that already benefit from agglomeration forces to build their own dynamics. It would
allow lagging regions to improve their capacity to absorb investment in innovation activity (Oughton
et al., 2002).
A related policy implication is the need for monitoring tools. The definition of regional policies
adapted to each regional context requires a good understanding of local characteristics. This relies on
economic studies to identify the characteristics of the region. The efficiency of local policies is based




h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
24

on a good understanding of scientific and technological potential, industrial structures, distribution
structures, local demand features, local channels of knowledge diffusion, and other factors.
Benchmarking with national and European regions would also provide useful information on the
strengths and weaknesses of each territory. Such benchmarking is also required at the level of the
innovation policy itself. Transnational policy learning is indeed important to identify best practices.
This will provide policy makers with an evaluation of what works and what does not as well as with
a clear idea of policy needs and existing arrangements, including public programs currently in place.

The relevant objectives and means defined above for innovation policies at the regional level are
summarised in Table 1. This paper does not propose that uniform regional policies should be
constructed. On the contrary, our results provide incentives for each region to be innovative in its
policy design and to find the policy mix that best corresponds to its own strategic needs.


Table 1
Synthesis of objectives and means for regional innovation policy
Means
Objective 1:

Good
understanding
of the local
characteristics

Economic studies to identify what characterizes the region: efficiency of local
policies is based on a good understanding of scientific and technological potential,
industrial structures, distribution structures, local demand features local channels
of knowledge diffusion.

Benchmarking with national and European regions.

Objective 2:

Trans-
National Policy
learning

Evaluation of past practices: necessity to know "what works" and "what does not", to
have a clear idea of needs and existing arrangements including public programs
currently in place.

Policy benchmarking at a national and European level

Objective 3:

Fostering pure
knowledge
spillovers

Supporting local academic research and publications (open science)

Designing local scientific and technological poles of excellence

Support to patenting by the local firms and universities

Facilitating access by local firms to the information contained in local, national and
international patents: information, advice, tax incentives for license purchasing.

Local platforms that foster the dissemination of knowledge. These include
transferring the results of public research to SMEs in an understandable form

Retaining Europe's academics within the ERA and reaping the benefits of knowledge
flows within Europe, by improving academic salaries and research opportunities.
Objective 4:

Science-
industry
knowledge
flows

(Transformation
of scientific
results into

Promoting entrepreneurial culture within the scientific field. "Education for
entrepreneurship" addresses not only students, but also researchers.

Financial, technical and informational support to diverse and complementary forms
of interaction, from simple transfer to complete cooperation.

- Informal links: seminars, consultations or visits to laboratories (these informal
links are a good way to generate opportunities).

h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
25

competitive
performances)

- Formal relationships within a bottom-up approach: contractual arrangements
with various lengths of time, supply of advice or training to companies, private
funding, research shared between public and industrial laboratories, "hiring"
students, creation of incubators by university laboratories ("academic
incubators"), cooperative/joint research, joint research centers.


Objective 5:

Inter and intra
firm
knowledge
flows

(Setting up a
regional
comparative
advantage)

Financial, technical and informational supports to local multidisciplinary cooperative
structures generating collective innovation processes.

In accordance with local industrial structures, the promotion of local transverse
cooperation programs (encouraging trans-sector-based and user-driven cooperative
structures and on a wider level, meetings on a transverse theme).

Supporting access for local businesses (mainly SMEs) to the latest methods of
knowledge management: seminars, financing specialized technical consultants.

Objective 6:

To connect the
local
innovation
system to
international
levels

(Territorial
absorptive
capacity and
attractiveness)

Promoting a regional learning economy

Physical infrastructures, equipment and human resources for the education system.
Higher education institutions. A key role assigned to University.

Education and training policy, developing ways to learn, stimulating a culture to
learn (for example rewards depending on learning and creative efforts)


Developing communication infrastructures

Development of communication infrastructures in all their forms.
Diversifying means of communication.


Supporting access for local businesses to European programs

Advisory activities for SMEs in order to allow them to join the European cooperative
structures


Acknowledgements

This research has received funding from the European Community's Seventh Framework
Programme (FP7/2007-2013) under grant agreement n 216813. The paper has benefited from the
discussions associated with the IAREG European program. We are especially grateful to Gerd
Schienstock, Edward Bergman, Stefano Usai and Simona Iammarino for suggestions and helpful
comments. None of these, however, are responsible for any remaining errors.


References

Almeida P., Kogut B. (1999), " Localization of knowledge and the mobility of engineers in regional networks
", Management Science, 45, 905-917.
Alvarez I., Molero J. (2005), Technology and the generation of international knowledge spillovers: An
application to Spanish manufacturing firms, Research Policy, 34, 14401452.Amiti M. (1998), "New
trade theory and industrial location in the EU: a survey of evidence", Oxford Review Economic Policy,
vol.14, n2.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
26

Amin and Cohendet (2004), Architectures of Knowledge: Firms, Capabilities and Communities, Oxford
University Press, 196.
Antonelli C. (1999), "The Evolution of the Industrial Organisation of the Production of Knowledge ",
Cambridge Journal of Economics, 23, pp. 243-260.
Antonelli C. (2001), "Learning to communicate in the production of collective knowledge", In Feldman M.,
Massard N., Institutions and systems in the geography of innovation, Kluwer Academic Publishers,
Boston, Dordrecht, London, 21-33.
Arrow K. (1962), "Economic Welfare and the Allocation of Ressources for Invention". in The Rate and
Direction of Inventive Activity : Economic and Social Factors, R. Nelson, NBER, Princeton University
Press : 164-181.

Audretsch D. B. (1995), Innovation and Industry Evolution. Cambridge, MA: MIT Press. 221.
Audretsch D. and Feldman M. (1996), R&D Spillovers and the Geography of Innovation and Production, The
American Economic Review, vol. 86, no. 3, jun, p. 630-640.
Audretsch D. and Feldman M. (1999) Innovation in cities: science-based diversity, specialization and
localized competition, European Economic Review 43, p. 409-429.
Audretsch D.B. and Keilbach M. (2004) Entrepreneurship Capital and Economic Performance. Regional
Studies, Vol. 38, p. 949.
Autant-Bernard C. (2001), Science and Knowledge Flows: Evidence from the French Case. Research Policy
30, 1069-78.
Autant-Bernard, Guironnet J.P., Massard N. (2009) Productivity Changes and Intangible Assets: Evidences
from French Plants (2002-2002), IAREG Working Paper 4.5.
Autant-Bernard C., LeSage J.P. (2009) Quantifying knowledge spillovers using spatial econometric tools,
Working paper.
Autant-Bernard, Massard N. (2009) Report on the underlying mechanisms of knowledge diffusion, IAREG
deliverable 4.2b.
Bala V., Goyal S. (2000) A non-cooperative model of network formation. Econometrica 68: 11811229
Bekkers R., Bodas Freitas I.M (2008) Analysing knowledge transfer channels between universities and
industry: To what degree do sectors also matter?, Research Policy 37, 1837-1853.
Bergman E. (2009), "Hirschman Faculties: Brain Circulation and Knowledge Flows of European University
Academics" IAREG Working paper 4.2
Bergman E., Usai S. (2009), "Knowledge Diffusion in European Regions", IAREG Delivrable 4.1
Bernstein J., Nadiri I. (1989) "Research and Development and Intraindustry Spillovers: An Empirical
Application of Dynamic Duality," NBER Working Papers.
Boschma, R. (2005), "Editorial: Role of Proximity in Interaction and Performance: Conceptual and Empirical
Challenges," Regional Studies, Taylor and Francis Journals, vol. 39(1), p. 41-45, January.
Bottazzi L., Peri G. (2003), Innovation and spillovers in regions: Evidence from European patent data,
European Economic Review 47(4), 687-710.
Caracostas, P., Soete, L. (1997), "The building of cross-border institutions in Europe: towards a European
system of innovation", In Edquist, C., Systems of innovation, Pinter, London and Washington, 395-419.
Carlsson, B., Jacobsson, S. (1997), "Diversity creation and technological systems: a technology policy
perspective", In Edquist, C., Systems of innovation, Pinter, London and Washington, 266-294.
Cohen W., Levinthal D. (1989), "Innovations and Learning : The Two Faces of R-D", The Economic Journal,

99, Septembre, p. 569-596.
Cohendet, P., Llerena, P. (1997), "Learning, technical change and public policy: how to create and exploit
diversity", In Edquist, C., Systems of innovation, Pinter, London and Washington, 223-241.
Conti,A., P. Gaul, and D. Foray (2007) Academic Licensing: a European Study CDM Working Papers
Series.
Cowan, R., Foray, D. (1997), "The economics of codification and the diffusion of knowledge", Industrial and
corporate change, 6, 595-622.
Cowan R., Zinovyeva N. (2009) "Papers or patents: Channels of University effect on regional innovation ".
Working Paper, p.36.
Cohen W.M., Nelson R.R., Walsh J.P. (2002) Links and Impacts: The Influence of Public Research on
Industrial R&D, Management Science 48 (1), 1-23.
Dasgupta, P., David, P. (1994), "Towards a new economics of science", Research Policy, 23, 487-521.
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
27

David P., Foray D. (1994), Accessing and expanding the science and technology knowledge-base: a
framework for interpreting available quantitative measures, OCDE-DSTI-CSPT.
David P., Foray D. (2002), Economic Fundamentals of the Knowledge Society, CEMI-REPORT.
DEste, Iammarino S. (2009) The spatial profile of university-business research Partnerships, IAREG
Working Paper 1.3a.
De Groot H., Poot J. and Smit M. (2007) Agglomeration, Innovation and Regional Development: Theoretical
Perspectives and Meta-Analysis, Tinbergen Institute Discussion Paper.
Dosi, G. (1988), Sources, procedures and micoeconomic effects of innovation, Journal of Economic
Literature, XXVI, 1120-71.
Dosi, G., Llerena, P. and Syslos-Labini, M. (2006) The relation between Science, Technology and their
industrial exploitation: an illustration through the myths and realities of the
so-called European Paradox Research Policy, Vol 35, n10, pp. 1450-1464.
Ernberg, E., Jacobsson, S. (1997), "Technological discontinuities and incumbents performance: an analytical
framework", In Edquist, C., Systems of innovation, Pinter, London and Washington, 318-341.
European Commission, (2007) ERIK Report, European Regions Knowledge based Innovation Network ,
April.
Fadairo, M. (2002), "Technology externalities and compromise in innovation policy; the European Union
case", In Feldman M., Massard N., Institutions and systems in the geography of innovation, Kluwer
Academic Publishers, Boston, Dordrecht, London, 293-308.
Fadairo, M., Massard, N. (2000), "La dimension territoriale des externalits de connaissance", In Delapierre,
M., Moati, P., Mouhoud, M., Connaissances et Mondialisation, Economica, 31-39.
Feldman, M., Massard, N. (2002), Institutions and systems in the geography of innovation, Kluwer Publisher,
384 p.
Foray, D. (1997), Generation And Distribution Of Technological Knowledge : Incentives, Norms And
Institutions , In Edquist C., Systems Of Innovation, Pinter, London And Washington, 64-85.
Foray, D., Mairesse, J. (2002), "The knowledge dilemma and the geography of innovation", In Feldman M.,
Massard N., Institutions and systems in the geography of innovation, Kluwer Academic Publishers,
Boston, Dordrecht, London, 35-54.
Foray, D., David, P.A. and Hall, B. (2009), Smart Specialisation. The concept, Knowledge Economists
Policy Brief, n9, Expert group on Knowledge for growth.
Giuri and Mariani, 2007 Giuri, P., Mariani M., 2007, Proximity of Inventors and Knowledge Flows,
Mimeo.Hageman A., Kelly M. (1999), " Marshallian externalities in innovation" Journal of economic
growth, 4, 39-54.
Glaeser E. L., Kallal H. D., Scheikman J. A., Shleifer A. (1992), " Growth in cities ", Journal of Political
Economy, vol. 100, n 6, p. 1126-1152.
Gomes-Casseres B., Hagedoorn J., Jaffe A. (2006) Do alliances promote knowledge flows?, Journal of
Financial Economics, 80:1, pp. 5-33.Goto, A., Suzuki, K. (1989), R&D capital, rate of return on R&D
investment and spillover of R&D in Japanese manufacturing industries. Review of Economics and
Statistics 71, 555564.
Granovetter M. (1985) Economic action and social structure: the problem of embeddedness. American Journal
of Sociology 91: 481-510.Hirschmann A.O. (1958), The Strategy of Economic Development. New
Haven, Conn.: Yale University Press.
Howells J. (2005), "Innovation and regional economic development: A matter of perspective?", Research
Policy, 34, 12201234.
Iammarino S., Piva M., Vivarelli M., von Tunzelmann N. (2010) Technological capabilities and patterns of
cooperation of UK firms: A regional investigation, IAREG Working paper, 1.2a.
Jaffe A.B. (1986) Technology Opportunity and Spillovers of R&D: Evidence from Firms Patents, Profits,
Market Value. American Economic Review 76, 984-1001.
Jaffe A.B. (1989), " Real effects of academic research ", The American Economic Review, vol. 79, n 5,
dcembre, p. 957-970.
Jaffe A.B., Trajtenberg M., Henderson R. (1993), " Geographic localization of knowledge spillovers as
evidenced by patent citations ", The Quaterly Journal of Economics, August, p. 577-598.
Jackson M., Wolinsky A., (1996) A strategic model of economic and social networks, Journal of Economic
Theory 71: 4474
h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0
28

Johnson, B. (1992 ), "Towards a new approach to national systems of innovation", In B.-A. Lundvall,
National systems of innovation, Pinter, London and New York, 23-44.
Joly, P.-B. (1992), "Le rle des externalits dans les systmes dinnovation", Revue Economique, No. 4, July,
785-796.
Laranjaa, M., Uyarrab, E., Flanaganb K. (2008), " Policies for science, technology and innovation: Translating
rationales into regional policies in a multi-level setting ", Research Policy, 37, 823835.
Lucas R. (1988) On the mechanics of economic development, Journal of Monetary Economics, 22, pp. 3-42.
Lundval, B.-A. (1992), National Systems of Innovation, Pinter 342.
Lundval, B.-A. (1998), "The learning economy: challenges to economic theory and policy", In Nielsen, K.,
Johnson, B., Institutions and Economic Change, Edward Elgar, 33-54.
Lundquist K-J., Trippl M. (2009) Towards cross-border innovation spaces, IAREG Working Papers 4.1.
Maggioni, M. (2001), "The development of high tech clusters: theoretical insights and policy implications", In
Feldman, M., Massard, N., Institutions and systems in the geography of innovation, Kluwer Academic
Publishers, Boston, Dordrecht, London, 309-339.
Maskell, P. (2001) "Future challenges and institutional preconditions for regional development policy", in
Metcalfe, J.S. (1994), "Evolutionary economics and technology policy", The Economic Journal, 104, July,
931-944.
Mohnen, P., Nadiri, M.I., Prucha, I. (1986) R&D, production structure and rates of returns in the US, Japanese
and German manufacturing sectors. European Economic Review 30, 749771.
Morgan, K., (2004) "The exaggerated death of geography: learning, proximity and territorial innovation
systems", Journal of Economic Geography, 4: 321.
Nauwelaers C., Wintjes R., (2008), "Innovation policy, innovation in policy: policy learning within and across
systems and clusters in Nauwelaers C., Wintjes R., Eds., Innovation Policy in Europe, Measurement
and Strategy, Edward Elgar, 225-268.
OMahony, M., Vecchi, M. (2009) R&D, Knowledge Spillovers and Company Productivity Performance.
Research Policy 38, 35-44.
OECD (2009), Public Governance and Territorial Development Directorate, Territorial Development Policy
Committee, Report GOV/TDPC/MIN(2009)2/REV1, March.
Oughton C., Landabaso M., Morgan K. (2002) The regional innovation paradox: innovation policy and
industrial policy, Journal of Technology Transfer, 27, 97-110.
Paci R., Batteta. (2003, "Innovation Networks and Knowledge Flows across the European Regions," Working
Paper CRENoS 200313
Peri G. (2005) Determinants of Knowledge Flows and Their Effect on Innovation. The Review of Economics
and Statistics 87 (2), 308-322.
Rallet A., Torre A. (2007), Quelles proximits pour innover ?, LHarmattan, 221.
Sena V. (2004) Total factor productivity and the spillover hypothesis: Some new evidence, International
Journal of Production Economics 92, 31-42.
Schienstock G. (2009), Knowledge Management Practices in Low-tech and Medium-tech Industries. Findings
from a Finnish Business Survey, IAREG Working Paper 4.3.
Schiller and Revilla-Diez (2009) Mobile star scientists as regional knowledge spillover agents, IAREG
Working Paper 2.7.
Singh J. (2005) Collaboration networks as determinants of knowledge diffusion patterns, Management
Science, 51 (5), 756-770.
Sorenson O., Rivkin J.W., Fleming L. (2006) Complexity, networks and knowledge flow, Research Policy,
35, 2006, 994-1017.
Tdtling, F., Trippl M. (2005), "One size fits all? Towards a differentiated regional innovation policy
approach", Research Policy, 34, 12031219.
Varga A., Parag (2009) Academic Knowledge Transfers and the Structure of International Research
Networks, IAREG Working Paper 1.3d.
Zucker L.G., Darby M.R. (2007), Virtuous circles in science and commerce, Papers in Regional Science, 86
(3) August,pp 445470
Zucker L.G., Darby M.R., Armstrong J. (1994), Intellectual Capital and the Firm : the
Technology of Geographically Localized Knowledge Spillovers, NBER Working Paper Series,
Working Paper, no. 4946, NBER, Cambridge, p. 59.

h
a
l
s
h
s
-
0
0
4
9
1
0
6
2
,

v
e
r
s
i
o
n

1

-

1
0

J
u
n

2
0
1
0